• derpgon@programming.dev
    link
    fedilink
    arrow-up
    36
    ·
    10 months ago

    The circle of life continues, and literacy goes down. AI cannot proofread, it merely says “these letters usually go with these”. AI screws up, people get taught shit language, they use it, it gets used as training data, rinse and repeat.

    • Adderbox76@lemmy.ca
      link
      fedilink
      English
      arrow-up
      14
      ·
      10 months ago

      The Extinction level meteor can’t come soon enough.

      Time to pack it in and give some other microorganism a shot at the evolutionary big-leagues. Maybe they’ll do better.

      • Appoxo@lemmy.dbzer0.com
        link
        fedilink
        arrow-up
        3
        ·
        10 months ago

        Assuming we take the Halo lore as a matter of fact and are not projecting human faults to the Covenant: I guess not really.

        As soon as monarchs or religious leaders emerge it’s game over.
        And I’d guess there will always be someone more rich than the other one and be in charge of something more resulting in something like a capitalist system.

    • Zworf@beehaw.org
      link
      fedilink
      arrow-up
      7
      ·
      edit-2
      10 months ago

      Well language is a fluid thing. If more and more people get taught shit language, the language will change to match. We have far worse problems on this planet :)

      I do think these AI companies grabbing what they can without giving anything back is a problem though. In my view content creators are a bit hypocritical too though. When Google scraped the internet verbatim (viewable in google cache) they didn’t mind because it gave them discoverability. Now they suddenly do care because they don’t benefit directly. Really, the stance should have been made earlier. But I do agree it should be stopped. Or content creators compensated.

    • wooki@lemmynsfw.com
      link
      fedilink
      English
      arrow-up
      1
      ·
      10 months ago

      Absolutely the large language models are over glorified word predictors that get it wrong. I’d go so far as to say, they get it wrong nearly all the time.